Arsenic and fluorine in groundwater in northern Mexico: spatial distribution and enrichment factors

North-central Mexico has groundwater contaminated with arsenic (As) and fluoride (F). Based on the dispersion patterns of these solutes, their sources are linked to felsic volcanic rock fragments and secondary minerals (clays, iron oxyhydroxides) within the alluvium fill of the aquifers. However, little is known about the effect of the enrichment factors for F and As in this area. Natural enrichment factors include evaporation, Ca/Na, and competitive adsorption and desorption from solid phases. This study used 1237 groundwater quality data measurements from 305 sampling sites collected between 2012 and 2019 in the state of Durango in north-central Mexico. To determine the contribution of enrichment factors to As and F content, the study area was divided into four sections, two being in the mountainous part of the state and two in the high plateaus. The data were compared among sections and analyzed using Spearman correlation and Piper and Block diagrams. The results indicate that the main solute enrichment mechanisms are evaporation and weathering of silicates and evaporites. Among the four sections, As, pH, and HCO3 seemed not to vary, F varied slightly, and nitrate and total dissolved solids varied the most. The lack of variation in As among sections is associated to its strong adsorption to clay minerals and iron oxyhydroxides, whereas the diminished F content in the eastern sections is likely linked to the adsorption of F to precipitating calcite (since groundwater is saturated with respect to calcite (SIcalcite = 0.43) and undersaturated for fluorite (SIfluorite =  − 1.16). These processes shed light on the distribution of F and As in this area, and are likely operating in other states in northern Mexico and in semi-arid areas elsewhere.

in arid and semi-arid areas such as the Chacopampean region of Argentina (García et al., 2014), northern China (He et al., 2020), and the north-central part of Mexico (González-Horta et al., 2015;Navarro et al., 2017;Alarcón-Herrera et al., 2020;Gutiérrez et al., 2021a). Groundwater with high As and F concentrations has also been reported in areas of recent volcanism, such as the Ethiopian Rift (Bianchini et al., 2020) and central Italy (Cinti et al., 2019).
The north-central portion of Mexico has high As and F concentrations, with many wells surpassing the recommended guidelines to safe drinking water, exposing a total of 8.8 million people to high As and 3 million to high F (Alarcón-Herrera et al., 2020). These elements have been reported geogenic in origin and tend to co-occur (Reyes-Gómez et al., 2013). Also, depending on their concentration in drinking water, they can be toxic for human consumption (González-Horta et al., 2015;Ortiz-Letechipia et al., 2022, Reyes-Gómez et al., 2013. However, more information about the correlation between As-F concentrations, the materials responsible for releasing these ions to groundwater and the factors causing enrichment are needed to better understand the spatial concentration patterns observed. The enrichment factors commonly reported as affecting As and F concentrations include groundwater withdrawals, evaporation, salinity, and competitive adsorption. The extent to which each of these processes affects the As and F concentrations is key to identify the wells and conditions that assure safe drinking water and to devise water treatments for the affected wells. The study was focused within the state of Durango, Mexico, because it is at the center of a larger area where high As and F concentrations have been reported and there is a good coverage of data. The state of Durango has similar climate, elevation, and geology than other states in north-central Mexico (Chihuahua, Coahuila, Zacatecas, and San Luis Potosi). Therefore, the geohydrological processes that affect As-F correlation and enrichment factors operating in Durango are expected to be similar to those operating in these other states and in semi-arid areas elsewhere.
Various sources of As and F have been identified, among them soluble minerals that form from deposition of volcanic gases filling voids in the volcanic rocks (Ren et al., 2022;Rosenberg, 1988), weathering of rock (mainly felsic volcanic rock such as tuff, ignimbrite, and volcanic glass) (Cinti et al., 2019;Alarcón-Herrera and Gutiérrez, 2022;Nordstrom, 2022;Podgorsky & Berg, 2022), and desorption from clay minerals and iron oxyhydroxides (Kumar et al., 2020). Once As and F ions are released, they adsorb onto clay minerals and iron oxyhydroxides present in the alluvial fill depending on the adsorption sites available, pH, and presence of competing ions (Rathore et al., 2016;Scanlon et al., 2009). Arsenic adsorption is less dependent of pH than F, and both of them may desorb under favored conditions on redox, pH, and ionic content (Kumar et al., 2020;Rathore et al., 2016). Near the study area, an association between As, F, and felsic volcanic rocks has been reported for Texas (Scanlon et al., 2009), Chihuahua (Gutiérrez et al., 2021a;Reyes-Gómez et al., 2013), Zacatecas (Ortiz Letechipia et al., 2022, and Jalisco (Ortega-Guerrero, 2009). An additional source of F is the dissolution of fluorite (CaF 2 ) that forms when hydrothermal fluids rich in F become in contact with Ca-carbonate rocks (González-Partida et al., 2019).
Both As and F are toxic at small concentrations. Chronic ingestion of these contaminants produces arsenicosis and fluorosis, respectively. Health effects are beyond the scope of this study; however, recommended guidelines related to human health and studies reporting their toxicity and health risks are listed in Table 1.
The objectives of this study are as follows: 1. Determine the As and F content and spatial distribution in each of the four sections comprising the study area and compare the As and F content among sections to determine the relative impact of As and F content according to the different geomorphological and geological characteristics (mountainous or high plateau) that characterize each section. 2. Identify possible relationships of As and F content with various enrichment factors, including its location in wither a closed or open basin, evaporation, lithology, and content of Na + , HCO 3 − , Ca 2+ , and TDS.

Description of the study area
The state of Durango in northern Mexico is bound in its western part by the Sierra Madre Occidental and has high plateaus in the eastern part. The Sierra Madre Occidental provides a rain shadow that drops the precipitation from 600 mm in its western margin to 200 mm in its eastern part. Most of the surface on the state varies in elevation between 1100 and 2700 m. Its physiography is mainly described as mountains or high plateaus (CONAGUA, 2009).
The outcropping rocks in the state comprise volcanic rock in the western and central part and sedimentary rocks of marine origin in its eastern part (Fig. 1). The low elevation part of the state (about 39% of the surface area) is covered by a thick layer of alluvium. The major components of the alluvium are 61% volcanic rock (mainly tuff and rhyolite), 16% sedimentary tock (limestone and sandstone), and 20% fine-grained material (CONAGUA, 2009).
The Sierra Madre Occidental formed after several episodes of volcanic eruptions of felsic magma throughout the Tertiary Period. This explosive type of volcanism extrudes large amounts of pyroclastic material and volcanic gases (Ferrari et al., 2007) that are rich in As and F. Both As and F are incompatible during the crystallization of mafic magmas and therefore become enriched in the melt and volatile components during the late stage (felsic) of magma differentiation (Nordstrom, 2022). Because of the rapid cooling, volcanic gases become trapped in the rock matrix and form a variety of minerals depending on their composition and temperature of formation (Rosenberg, 1988;Wallace, 2010). Some minerals that form after sublimation and deposition of volcanic gases include the soluble arsenogoyazitearsenoflorencite group (Ren et al., 2022) and aluminum fluoride hydrates (Rosenberg, 1988). Besides F and As contained in the felsic rocks, volcanic glasses, and hydrated salt depositions, magmatic fluids may crystallize as CaF 2 (fluorite) deposits, some of which occur in the central part of the study area (see Fig. 1). Mexico's largest fluorite mineralizations outcrop south and north of the study area, in San Luis Potosí and Coahuila, respectively (González-Partida et al., 2019).

Data collection and processing
Data were obtained from Mexico's government water agency CONAGUA website for the years between 2012 and 2019. Information about the analytical methods for each parameter and their quality control and assurance is presented in the National Network for Water Quality (RENAMECA). A table of the analytical methods used in the CONAGUA National Laboratory is provided in the supplemental information (Table S1). The ionic balance was < 5% error for the majority of the samples and had a median value of 4.74%. Saturation indices (SI) were calculated using Visual Minteq software.
The data were screened for entries containing all parameters of interest (As, F, HCO 3 , Ca, Na, and TDS). CONAGUA sampled the wells from its monitoring network once or two times a year, although some wells may not be sampled during 1 particular year. A total of 1237 sample points collected between 2012 and 2019 from 305 sampling sites were obtained. The distribution of sampling sites is shown in Fig. 2. Due to the significantly different topography and geology within the study area, the data were divided into four sections according to Table 1 Health problems associated with ingestion of water containing As or F, and recommended guidelines for the WHO and selected countries

Element Health concerns Guidelines References
As Skin lesions; chronic ingestion causes more severe health problems such as nervous and certain types of cancer. Food may be an important As source in some countries their location (NE, NW, SE, SW) to see if there is a relationship between the As and F content and to either its proximity to the Sierra Madre Occidental (sections NW, SW) and to the Comarca Lagunera, a large agricultural area (sections NE and SE), as well as the potential effect of several enrichment factors reported by other studies. The resulting four sections are of roughly the same size, using 25° latitude to separate north from south; the northern part was separated by meridian 104° 45′ W, and the southern section by meridian 104° 15′ (Fig. 2). In all four sections, the aquifers are composed of granular material (alluvium) with a thickness varying between 100 and 1000 m, and mainly open except for places where clay lenses are present, and the aquifer behaves as a confined aquifer. Surface water and groundwater flow in a west to east general direction until reaching the closed basins in the eastern part of Durango, where the water either accumulates or evaporates.

Maps, diagrams, and statistical analyses
A Piper diagram was constructed to determine the changes in main solutes for the different sections within the study area. The GW_Chart software (USGS) was used to construct the Piper diagram. All other diagrams were constructed using MS Excel (version 2017). The effect of the main lithology onto water chemistry was inferred via a Na-normalized Ca vs HCO 3 diagram (after Gaillardet et al., 1999), and Boxplot diagrams were constructed for each of As and F data, according to their location in either a closed basin or open basin. The Boxplot diagrams contain half of the data within the box with a median value represented as a line within the box, and data outside the whiskers representing anomalies.
Maps were constructed using ArcMap 10.8, and a 2013 USA Topographic map as base map. The projection of the map is GCS Mexico ITRF2008, with the scale set at 1:2,950,000. Correlations between As and F as well as the enrichment factor were determined in MS Excel. The non-parametric Spearman correlation was utilized as the data are not normally distributed. For statistical purposes, below detection values were substituted for one-half of the detection limit value. The significance of ρ was determined using a two-tailed Student's t-test with n-2 degrees of freedom (Frost, 2019).

Results and discussion
High As (> 0.025 mg/L) and F (> 1.5 mg/L) were observed in 50.2% and 31.1% of the sampled sites, respectively. The As and F concentrations (Fig. 3) plot in a dispersed pattern over the study area, but low concentrations follow this pattern too, implying that chemical weathering of the alluvium fill is an important source of As and F and that fluorite mineral deposits (their location is shown in Fig. 1) may also contribute F to groundwater to a lesser extent.
The concentrations (mean and standard deviation) of As, F, associated solutes, and the Spearman correlation coefficient ρ (As-F) are listed in Table 1 for each of the four sections within the study area. The composition of alluvium was assumed to be similar throughout the study area based on the predominant presence of only two main types of outcropping rocks: felsic volcanic and calcium carbonate rocks (see Fig. 1).
Evaporation is a factor associated with enrichment of As and F (Alarcón-Herrera et al., 2020; Feng et al., 2022). However, more intense evaporation in the drier eastern sections did not seem to affect the As content proportionately, as shown in Table 2, probably because the whole study area falls under semi-arid climate and also as a result of the strong preference of As to adsorb onto clay minerals and iron oxyhydroxides. Similarly, F concentrations among sections would not be affected by evaporation since the conditions are the same as As. However, the variation of F among sections indicates that other enrichment factors affecting F may be at play.
Variations in F content among sections include higher concentration in the SW section and a decrease in concentration in the NE and SE sections downstream from their NW and SW sections, respectively. The high F concentration in the SW section suggests that the large fluorite mineral deposit nearby (González-Partida et al., 2019) may be contributing F to groundwater. The reduction of Ca/Na in the western sections ( Table 2) is indicative of fluorite dissolution (McMahon et al., 2020;Nordstrom, 2022) and a release of F. However, the smaller F concentration observed in the eastern sections indicates the presence of an F removal mechanism, such as adsorption onto precipitating calcite (Deng et al., 2016;Zulueta-Lacson et al., 2022). To test this possibility, saturation indices (SI) from each section were calculated for 8 representative samples per section. The results (Table S2) show oversaturation conditions for calcite (positive SI value) in 75% of the samples in both NW and SW sections and in 100% saturation conditions in the samples within the NE (average SI calcite = 0.43) and SE sections (SI calcite = 0.42). In contrast, SI for fluorite was negative in all of them, indicating conditions of undersaturation and an unlikely removal of F by crystallization of fluorite. However, adsorption of F in precipitating calcite is a viable process (Puccia et al., 2018; Turner With respect to TDS as an enrichment factor, limestone and other sedimentary rocks that outcrop in the eastern sections provide dissolved Ca, Mg, HCO 3 , and SO 4 . Their increase is likely exacerbated by both evaporation and land use, as they show the largest increase in the NE section, where intensive agriculture and dairy farms generate large amounts of waste (e.g., salt-rich drain returns and excess fertilizers) (Gutiérrez et al., 2021b). Since the sections high in TDS are not necessarily rich in As or F (Table 2), then TDS is not an As, F enrichment factor per se in the region. Precipitating calcite can serve as a seed to adsorb solutes (Deng et al., 2016), a process that is highly dependent on pH and favored at pH < 7 (Vital et al., 2019;Zulueta-Lacson et al., 2022). Such conditions are met in some wells of NE section, as shown in Table 2.
Additional insight to the As and F distribution was obtained from their associated ions. The changes in groundwater chemistry are depicted in a Piper diagram. To this purpose, eight water samples from each section were chosen at random and plotted. The results ( Fig. 4 and Table S2) show water of each section grouping together and a clear separation between sections. The values of one section compared with the values of the section downstream, e.g., NW to NE and SW to SE, plotted each in a different region within the diamond following an increase in SO 4 , Ca, and Mg concentrations.
The data were also plotted in a Ca/Na vs. HCO 3 / Na diagram to tie solutes to their lithological source (Gaillardet et al., 1999). A diagram was constructed for all samples in each section (Fig. 5). The results show a strong influence of silicate rocks in all four sections. Silicates include minerals present in the volcanic rocks plus clay minerals produced as a result of chemical weathering. The data plot tight along the weathering line for the mountainous NW and SW sections and more dispersed towards evaporite and carbonate for the high plateaus. Although all four sections show influence of both evaporites and carbonates in addition to silicate rocks, the effect is magnified in the high plateau areas. This magnification can be the result of furthering chemical weathering reactions, contact with caliche layers, or by the input of anthropogenically contaminated water.
Boxplot diagrams were constructed to compare the distribution of As and F values between open (exorheic) and closed (endorheic) basins. For this analysis, data were separated into two groups depending on their location, either within a closed basin or open basin. The results are shown in Fig. 6. These graphs confirm the non-normally distributed concentrations for As and F (the median is not located at the center of the box and there are numerous outliers).
The differences of F, As content among closed or open basins were smaller than it was expected. The As distribution shows both the medians (lines inside the box) slightly below the recommended limit of   (Table 1) varied between 0.259 (SE section) and 0.762 (SW section). This modest to strong As-F correlation agrees with values reported for surrounding areas that have similar rock outcrops and climate, e.g., ρ = 0.56 (Scanlon et al., 2009), r = 0.72 (Reyes-Gómez et al., 2013, and r = 0.71 (González-Horta et al., 2015).
A correlation between As, F, and solutes related to enrichment Ca/Na, HCO 3 , and TDS was carried out to look further into their associations. The data were normalized by Na to diminish the effect of anthropogenic contamination (Gaillardet et al., 1999). The results, summarized in Table 3, report a correlation similar to that obtained for non-normalized As and F values shown in Table 2, but with differences between them smoothed out. While As seems to not be affected by the presence of Ca and only slightly associated with HCO 3 , F association to these anthropogenic solutes is stronger. This difference in association between As and F with the other solutes may explain the overall modest  correlation obtained (ρ As-F = 0.44 to ρ As-F = 0.65) despite As and F originating from a common source. The results shed light onto the geochemical processes responsible for the variable content of As and F in central Mexico and explain their disseminated spatial pattern to a certain extent. However, the analysis was carried out over a large area, for which more detailed characterization of the lithology, pH, and water temperature, both laterally and with depth, would be required when studying a more constrained area.

Conclusions
High arsenic (> 0.025 mg/L) and fluoride (> 1.5 mg/L) were observed in 50.2% and 31.1% of the 1237 groundwater samples, respectively. The high As and F concentrations are plotted following a dispersed pattern over the study area, implying that their main source is the alluvium, which is comprised mostly by volcanic rock fragments derived from the Sierra Madre Occidental intermixed with fragments from older carbonate rocks. A significant difference in As and F concentrations was found among the four sections (NW, NE, SW, and SE) the study area was subdivided into, and differences were also found with respect to their water chemistry, with an increase of Ca and SO 4 in the eastern sections (plateaus). Among the sections, no appreciable changes were observed for pH, As, and HCO 3 ; F varied slightly; and NO 3 -N and TDS varied the most. The highest concentrations of salts were observed in the NE section, an expected result due to the intense agricultural activities taking place in this area. Surprisingly, boxplot diagrams revealed no significant differences between closed basins and open basins for As, and a slight but significant decrease of F in closed basins. The lesser F content observed in the NE and SE sections is likely the result of F adsorption onto precipitating calcite, since groundwater of closed basins was oversaturated with calcite and undersaturated with respect of fluorite.
The results emphasize a common origin of As and F, but once these solutes are released by the alluvium into groundwater, they are selectively enriched or depleted from the water by processes such as evaporation, adsorption, and coprecipitation. Although reported as major factors of enrichment, any changes in evaporation and alkalinity within the study area were not evident in the resulting water quality, probably because of the relatively similar climate and type of aquifer, and a mix of soluble rock, e.g., limestone fragments in the alluvium. Results obtained by the different methods of analyses (spatial distribution, Gaillardet diagrams, boxplot diagrams) were in agreement with the overall behavior of the solutes and together contributed to forming a more complete picture of the geochemical processes taking place in the area.